PKM2 preconditioning protects endothelial cells from pyroptosis and BBB disruption via NRF2/TRX/TXNIP signaling in neonatal hypoxic-ischemic brain injury

Abstract

Background

Neonatal hypoxic-ischemic brain damage (HIBD) is a leading cause of neurological deficits and death in neonates. In HIBD, the death of endothelial cells and disruption of the blood–brain barrier (BBB) are closely related to the severity of brain damage and long-term clinical outcomes. There is increasing evidence that a glycolytic enzyme, pyruvate kinase M2 (PKM2), is essential for managing metabolic processes in endothelial cells, but its role (and underlying molecular mechanism) in hypoxic-ischemic (HI)-associated endothelial cell metabolism, cell survival, and BBB function remains unknown.

Methods

We established an in vivo HI-induced brain injury rat model and an in vitro model in which human cerebral microvascular endothelial cells (hCMECs) underwent oxygen-glucose deprivation (OGD). Infarct volume was measured and neurobehavioral tests were conducted to assess brain damage, and Evans blue extravasation and FITC-dextran were used to evaluate the BBB. RNA sequencing, qRT–PCR, western blotting, and immunofluorescence labeling were conducted to identify the molecular mechanisms underlying HIBD.

Results

PKM2 expression was upregulated in the brains of HIBD rats and in OGD-treated hCMECs. The inhibition of PKM2 greatly upregulated the expression of pyroptosis-associated proteins, including NLRP3, cleaved caspase-1, GSDMD, IL-1β, and IL-18. In contrast, the activation of PKM2 preserved junctional proteins and maintained the integrity of the BBB, which together improved functional recovery in HIBD rats. Mechanistically, preconditioning of PKM2 contributed to lactate-mediated cellular defense mechanisms, including the activation of nuclear factor erythroid 2-related factor 2 (NRF2) and thioredoxin (TRX), and to the downregulation of thioredoxin-interacting protein (TXNIP) via a modest increase in reactive oxygen species.

Conclusions

Our analyses provide compelling evidence that PKM2 preconditioning attenuates endothelial cell pyroptosis and BBB disruption in neonatal HIBD by causing oxidative stress resistance and activating the NRF2/TRX/TXNIP pathway. Therefore, PKM2 represents a promising pharmacological target for treating HIBD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12974-026-03694-6.

Introduction

The induction of neonatal hypoxic-ischemic brain damage (HIBD) by asphyxia during the birthing process is a leading factor of cerebral damage in perinatal neonates, leading to cerebral palsy, seizures, cognitive impairment, behavioral anomalies, and even death [1, 2]. Current treatments for HIBD are limited to therapeutic hypothermia and the amelioration of resulting symptoms, which are not universally effective [3]. Mitochondrial damage, inflammation, oxidative stress, and autophagy contribute to poor outcomes following HIBD [4, 5], but the precise pathological mechanisms underlying this condition remain unclear.

The blood–brain barrier (BBB) helps to maintain homeostasis within the brain microenvironment, protecting neurons from environmental insults and promoting brain health and function [6, 7]. The BBB is comprised of brain endothelial cells connected by intercellular tight junctions, and their disruption has been linked to pathological changes in a number of neurological diseases, including cerebral hemorrhage, ischemic brain injury, and neonatal HIBD [8–10]. In HIBD, the hypoxic-ischemic (HI) microenvironment contributes to endothelial cell dysfunction and the downregulation of tight junction proteins, which results in disruption of the BBB [11]. Breakdown of the BBB allows immune cells and plasma proteins to infiltrate the brain parenchyma, which promotes inflammatory processes in the neonatal brain [6, 12]. These proinflammatory factors further disrupt the BBB.

Pyroptosis refers to a proinflammatory form of programmed cell death and depends on caspase-1 or caspase-11/4/5. During pyroptosis, the protein gasdermin D (GSDMD) is activated, which creates openings in the cell membrane that cause the leakage of cellular contents such as proinflammatory molecules [13, 14]. This form of cell death plays an important role in the progression of different neurological conditions, including stroke, traumatic brain injury, and HIBD [15–17]. In the context of HIBD, previous studies have focused on pyroptosis in neurons and glial cells [17, 18], but whether damage to cerebral vascular endothelial cells in HIBD results from pyroptosis remains unclear. Numerous studies have suggested that pyroptosis exacerbates the disruption of the BBB by promoting the secretion of inflammatory factors and reactive oxygen species (ROS), which cause irregular tight junctions and impair epithelial function [15, 19], but endothelial cell pyroptosis has not yet been demonstrated to compromise the integrity of the BBB in neonatal HIBD.

Glucose provides the main source of energy for the brain, which utilizes approximately 20% of total consumed glucose [20, 21]. Glucose metabolism in the brain generates ATP to provide energy via glycogen turnover, the pentose phosphate pathway, and glycolysis [22]. Pyruvate kinase (PK), a key enzyme in the glycolytic pathway, is encoded by two genes and exists in four subtypes: PKL, PKR, PKM1, and PKM2 [23]. The PKL isoform is found primarily in the kidneys and liver; PKR is found primarily in erythrocytes; PKM1 is distributed across many differentiated adult tissues, such as skeletal muscle, the myocardium, and brain tissue; and PKM2 is expressed during the growth period and in numerous adult tissues, such as the liver and brain [23, 24]. Over the past few years, the function of PKM2 in cancer and immune cells has received increasing attention [25, 26]. Studies have shown that PKM2 activation attenuates thoracic aortic aneurysm and dissection by blocking the release of IL-1β following the activation of NOD-like receptor family pyrin domain-containing 3 (NLRP3) [27]. Endothelial cells express high levels of PKM2, and it is currently thought that PKM2 plays a role in maintaining vascular integrity [28]. However, the effects of PKM2 and pyroptosis on endothelial cells following HIBD remain unclear.

This research focused on exploring the function of PKM2 in endothelial cells and clarifying the mechanism underlying HIBD. Using two HI models, namely, the Rice–Vannucci rat model and oxygen-glucose deprivation (OGD)-treated human cerebral microvascular endothelial cells (hCMECs), we sought to determine (1) whether HI-induced endothelial cell pyroptosis affected the integrity of the BBB, (2) whether PKM2 was involved in regulating endothelial cell metabolism and pyroptosis, and (3) whether PKM2 preconditioning could attenuate endothelial cell pyroptosis and BBB disruption and thereby ameliorate HIBD.

Materials and methods

Animals and animal welfare

The animal protocols for our experiments were authorized by the Animal Care and Use Committee of Wenzhou Medical University, China. Ten-week-old Sprague–Dawley (SD) rats of both genders were acquired from Zhejiang Vital River Experimental Animal Technology Co., Ltd. [SCXK (Zhe) 2023-0034]. Male and female rats were permitted to mate without restriction and were housed in a constant environment (50–60% humidity, 22 ± 2 °C, and a 12-hour light/dark cycle with the lights turned on at 8:00 AM) with unrestricted access to food and water. Each pregnant rat was housed separately. Male SD rats on postnatal day 7 (P7) were selected for in vivo experiments because of their lower baseline antioxidant capacity and higher susceptibility to HIBD relative to females [29, 30]. Sample size estimation was based on power calculations for the primary parameters (Morris water maze and infarct volume) to achieve a power of 0.8 and a significance level of 0.05 utilizing G*power software (v3.1) [31].

Neonatal HIBD model

The Rice–Vannucci protocol was adapted as outlined previously [32]. In brief, rats were anesthetized with isoflurane, and the left common carotid artery was permanently double-tied and severed to obstruct blood flow. After 2 h of recovery, the rat pups were placed in a hypoxia chamber containing 8% oxygen and 92% nitrogen at a constant temperature of 37 °C for 2.5 h. Rats in the sham group underwent artery isolation without ligation and were not exposed to hypoxic conditions. The mortality rate was 10% in the experimental group.

Experimental design

Individual SD rat pups were randomly allocated to one of the experimental groups using an online random number generator (https://www.graphpad.com/quickcalcs/randomize1/), and litter effects were controlled (max 1–2 pups per litter per group). Investigators were blinded to group allocation during the experiments, tissue processing, and image analysis.

Experiment I

To explore changes in pyroptosis and in the BBB in response to HI injury, the rats were randomly assigned to one of two groups: a sham group and an HI group. RNA sequencing, quantitative real-time PCR (qRT–PCR), western blotting, and evaluation of BBB permeability in the ipsilateral/left hemisphere were conducted 24 h after HI injury. To determine caspase-1 expression in endothelial cells, double immunofluorescence staining of caspase-1 with CD31 (an endothelial cell marker) was performed 24 h after HI injury in the HI group and in the sham group (n = 5/group).

Experiment II

To assess the effects of pyroptosis on BBB integrity, the rats were randomly assigned to one of five groups: sham, sham + nigericin, HI, HI + nigericin, or HI + Ac-YVAD. Either the NLRP3 activator nigericin (MedChemExpress, Catalog #HY-127019, 4 mg/kg/day) or the caspase-1 inhibitor Ac-YVAD (Sigma–Aldrich, Catalog #SML0429, 1 mg/kg/day) was injected intraperitoneally 2 h before HI injury. Western blotting, evaluation of BBB permeability, and double immunofluorescence staining to detect ZO-1 and CD31 expression were conducted 24 h after HI injury.

Experiment III

To evaluate PKM2 levels, samples were taken from the left/injured hemisphere (n = 5/time point) at 0 h, 6 h, 12 h, 24 h, and 48 h following HI injury. The protein concentration was detected using western blotting. Additionally, to assess PKM2 levels in endothelial cells, double immunofluorescence staining was performed to detect PKM2 and CD31 expression 24 h after HI injury in the HI group and in the sham group (n = 5/group).

Experiment IV

To determine the effects of PKM2 inhibition on pyroptosis, the PKM2 inhibitor Compound 3k (MedChemExpress, Catalog #HY-103617, 5 mg/kg/day) was administered intraperitoneally 2 h before HI injury, either alone or together with a 10 mg/kg/day dose of the NLRP3 inhibitor MCC950 (Sigma–Aldrich, Catalog #538120). Rats were assigned at random to either a sham, HI, HI + Compound 3k, or HI + Compound 3k + MCC950 group. Western blotting and double immunofluorescence staining were performed 24 h after HI injury.

Experiment V

To assess the effects of PKM2 preconditioning on short-term outcomes, long-term outcomes, and regulatory mechanisms, rats were assigned at random to either a sham, sham + ML265, HI, HI + ML265, HI + ML265 + ML385, or HI + ML265 + NAC group. Rats were intraperitoneally administered the PKM2 activator ML265 (Cayman Chemicals, Catalog #13942, 20 mg/kg/day) alone or together with either the nuclear factor erythroid 2-related factor 2 (NRF2) inhibitor ML385 (MedChemExpress, Catalog #HY-100523, 30 mg/kg/day) or the ROS scavenger N-acetylcysteine (NAC, MedChemExpress, Catalog #HY-B0215, 50 mg/kg/day) 2 h before HI injury and every 24 h thereafter. 2,3,5-Triphenyltetrazolium chloride (TTC) staining, Evans blue extravasation, western blotting, and double immunofluorescence staining were performed 24 h after HI injury. Hematoxylin and eosin (H&E; Solarbio Biotechnology, China) staining was performed 7 days after HI injury. Neurobehavioral tests were conducted 3 weeks after HI.

TTC staining

To visualize cerebral infarct areas, we used TTC staining as described previously [33]. Briefly, 24 h after HIBD, rat pups were placed under deep anesthesia with isoflurane and then perfused with 20 mL of cold saline. Isolated brain tissues were sliced into 2-mm-thick coronal sections and soaked in 1% TTC (Sigma–Aldrich, USA) in the dark for 30 min. The volume of the brain infarct was captured in photographs and quantified using ImageJ software. The percent infarct was determined using the following formula: [(total area of the contralateral hemisphere) − (area of the uninfarcted area of the ipsilateral hemisphere)] / (total area of the contralateral hemisphere × 2) for each slice [34].

Brain water content

To measure the water content of the brain tissue, the dry–wet ratio method was used to assess the degree of cerebral edema. After 24 h of HIBD modeling, the wet weight was determined by removing and weighing the left hemisphere (the injured hemisphere). Brain samples were dried in an oven at 65 °C for 48 h to determine their dry weight. The formula described by Tao et al. [35] was used to calculate the percentage of edema.

BBB permeability determination

Evans blue (Sigma–Aldrich, USA) extravasation was used to assess the permeability of the BBB, as described previously [10]. In brief, rats were injected with 4 mL/kg Evans blue solution (2% in phosphate-buffered saline) through the tail vein 24 h after HI injury. After 2 h, rats were placed under anesthesia and then perfused transcardially with precooled normal saline. Brains were extracted and weighed, brain tissue was homogenized in a 50% trichloroacetic acid solution, and the lysate was centrifuged at 12,000 rpm for 20 min at 4 °C. Afterward, the supernatant was collected and incubated with 70% absolute ethanol overnight at 4 °C. Following centrifugation, fluorescence signals were measured with a spectrophotometer at a wavelength of 620 nm. A standard curve was created to quantify the content of Evans blue dye in the brain tissue.

Morris water maze (MWM) test

The MWM test is a common neurobehavioral assessment used to measure spatial learning and memory capabilities, as described previously [36]. We performed the MWM test 3 weeks after HI injury. The tests were conducted in a quiet, dimly lit environment at the same time of day (14:00–18:00). The water maze featured a metal pool that was 140 cm wide and filled with water (temperature maintained at 22 ± 1 °C). The pool was made opaque by the addition of nontoxic black ink to obscure the submerged escape platform (15 cm in diameter), which was located approximately 2 cm below the water surface. The pool was divided into four equal quadrants, and the platform was placed in the southwest quadrant. During the training phase, each animal underwent four trials per day for five consecutive days. In each trial, the rats entered the water from the northwest, northeast, and southeast quadrants and were allowed to swim freely for 60 s to find the hidden platform. If the platform was not found within this time, the animal was guided to it and allowed to remain there for 15 s. The latency to reach the platform was recorded. Following the training phase, spatial acquisition testing was conducted on the sixth day. The rats were subjected to a 60-second spatial memory test at the same point at which the escape platform was removed. The frequency of crossings at the original platform location was measured to assess memory retention. The water maze activity was recorded and analyzed using a DigBehv system (Shanghai Jilang Software Technology Co., Ltd.). All the data were collected and processed without knowledge of the treatment conditions to maintain objectivity in evaluating the cognitive performance of the animals.

Histopathological and immunofluorescent evaluation of brain tissues

Excised brains were preserved in 4% paraformaldehyde at room temperature for one day and subsequently embedded in paraffin wax. After the paraffin was removed and the sections were rehydrated gradually, H&E solution was used to stain 5 μm paraffin sections according to standard methods. For double-immunofluorescence labeling, the sections were incubated with antibodies against CD31 (R&D Systems Catalog #AF3628, 1:100), caspase-1 (Abcam; Catalog #ab179515, 1:200), ZO-1 (Abcam; Catalog #ab221547, 1:100), MMP2 (Affinity; Catalog #AF5228, 1:200), NRF2 (Proteintech; Catalog #16396-1-AP, 1:200), 8-OHdG (Bioss; Catalog #bs-1278R, 1:200) or PKM2 (ZENBIO; Catalog #381318, 1:100) overnight at 4 °C. Then, the sections were washed, treated with Alexa Fluor 488 or Alexa Fluor 594 secondary antibodies for 1 h, and incubated with 4,6-diamidino-2-phenylindole (DAPI) for 10 min. A Nikon fluorescence microscope was used to capture images. Negative controls for immunofluorescence images are shown in Figure S9.

RNA sequencing

TRIzol reagent (Invitrogen, USA) was used to extract total RNA from the brain samples of rats in the sham and HI groups. An mRNA library was then created and sequenced using the BGIseq500 platform (BGI-Shenzhen, China). SOAPnuke was employed to filter the sequence data, as described previously [37]. Differentially expressed genes (DEGs) were analyzed using a threshold of p < 0.05 and |log₂FC| > 1. The significance levels of terms and pathways were corrected by using the Q value with a rigorous threshold (Q value < 0.05), according to the Bonferroni method.

Cell culture and treatment

HCMECs were obtained from Zhong Qiao Xin Zhou Biotechnology Co., Ltd. (Shanghai, China). Short tandem repeat markers of hCMECs were identified, and no contamination by mycoplasma was detected on a monthly basis. The cells were grown in endothelial cell medium (ECM, ScienCell, USA) supplemented with 5% fetal bovine serum, 1% penicillin/streptomycin, and 1% endothelial cell growth at 37 °C with 5% CO₂. HCMECs were subjected to OGD for 18 h of in vitro culture. Briefly, the complete culture medium was replaced with glucose-free culture medium, and the cells were cultured in a hypoxic incubator (1% O₂ / 5% CO₂) at 37 °C for 18 h without recovery, as described previously [38].

For the intracellular signaling assay, hCMECs were pretreated with nigericin (MedChemExpress, Catalog #HY-127019, 10 µM), MCC950 (Sigma–Aldrich, Catalog #538120, 10 µM), ML265 (Cayman Chemicals, Catalog #13942, 20 µM), ML385 (MedChemExpress, Catalog #HY-100523, 20 µM), galloflavin (MedChemExpress, Catalog #HY-W040118, 15 µM), oxamate (MedChemExpress, Catalog #HY-W013032A, 10 mM), lactate (Sigma–Aldrich, Catalog #71718, 10 mM) or control buffer for 2 h. The cells were subjected to OGD for 18 h.

To knockdown PKM2 expression by siRNA, hCMECs were cultured in six-well plates and grown to 60–70% confluence. Transient transfections were performed using Lipofectamine 3000 (Invitrogen, USA). After transfection with either control siRNA (RiboBio, Guangzhou, China) or PKM2 siRNA (5′-GGATGTTGATATGGTGTTT-3′, RiboBio, Guangzhou, China) for 6 h, the cells were treated as described above.

For overexpression of PKM2, hCMECs were transfected with the recombinant plasmid Flag-PKM2 FL (full length) or the pcDNA 3.1 empty vector with an Entranster-D4000 (Engreen Biosystem Co., Ltd., China).

Cell viability

To evaluate cell viability, the cell counting kit-8 (CCK-8) reagent (Yeasen Biotechnology, Shanghai, China) was used. Briefly, hCMECs were distributed in 96-well plates at a concentration of 10⁴ cells per well and subjected to different periods of OGD (0, 6, 12, 18, or 24 h). After 10 µL of CCK-8 solution was added to each well, the plates were maintained in the dark at 37 °C for 1 h. Optical density was measured at 450 nm, and the results were calculated as the percentage of live cells relative to that of the control group.

Cell death assay

An assay for measuring lactate dehydrogenase (LDH) release was used to evaluate pyroptotic cell death in rats and hCMECs. The rats were euthanized, and the left hemisphere of the brain was extracted. Tissue homogenates at a 10% concentration were prepared using extract liquid. HCMEC supernatants were also harvested and centrifuged (1000 rpm, 5 min) following OGD for 18 h. LDH activity was measured using a commercially available kit (BC0685; Solarbio Biotechnology, China).

FITC-dextran flux assay

FITC-dextran (Sigma–Aldrich, USA) was used to measure cell permeability, as previously reported [10]. HCMECs were plated in 24-well transwell chambers with 0.4-mm pores (Corning, USA). Following the experimental periods described above, the culture medium in the apical chamber was replaced with 200 µL of FITC-dextran medium (1 mg/mL), while the medium in the basal chamber was substituted with 300 µL of PBS. After a 1 h incubation period, the medium was collected for fluorescence measurement with a microplate reader (Thermo Fisher Scientific, USA) at an excitation wavelength of 492 nm and an emission wavelength of 520 nm. For in vivo experiments, rats were intravenously injected with 2 µL/g body weight of FITC-dextran (3 mg/mL). After 10 min of FITC-dextran circulation, rats were perfused with PBS at a rate of 10 mL/min. Tissue samples were weighed, homogenized with PBS, and centrifuged at 14,000 rpm for 10 min at 4 °C. FITC-dextran fluorescence was quantified as described above.

Lactate detection

After 18 h of OGD, the hCMECs were lysed ultrasonically. Cell supernatants were collected, and a commercial detection kit (Solarbio Biotechnology, China) was used to measure lactate levels.

Enzyme-linked immunosorbent assay (ELISA)

The level of IL-1β secretion from brain microvessel fractions was quantified using a rat ELISA kit (Jianglaibio, China) according to the manufacturer’s protocol.

ROS assay

An ROS assay kit (S0033; Beyotime) was used to evaluate intracellular ROS levels. HCMECs were exposed to 10 µM DCFH-DA at 37 °C for 20 min. Hoechst stain was used to stain the cell nuclei in the dark. Finally, live cells were imaged with a fluorescence microscope (Nikon, Japan), and the fluorescence brightness was evaluated with ImageJ software.

In vitro Immunofluorescence staining

Following a 15-minute exposure to 4% formaldehyde, hCMECs were incubated with 0.3% Triton X-100 in 10% BSA for 1 h. Subsequently, the same procedure as described above for tissue fluorescence was repeated on the cells.

qRT–PCR assay

Total RNA was extracted from brain tissues and hCMECs using TRIzol (Invitrogen, USA), and reverse transcription was performed with PrimeScript RT Master Mix (Takara, Japan). A CFX96 manager (Bio-Rad, Singapore) was used for qRT–PCR with TB Green Premix Ex Taq II (Takara, Japan). β-Actin was used as the internal control to normalize the expression of target genes. The gene-specific PCR primers used are listed in Table S2.

Western blotting and Immunoprecipitation

Using RIPA lysis buffer containing protease and phosphatase inhibitors (all from Solarbio Biotechnology, China), left/ipsilateral hemisphere tissue or hCMECs were homogenized. For nuclear extraction experiments, a nuclear extraction kit (R0050; Solarbio Biotechnology, China) was used to separate the cytoplasmic and nuclear fractions, and protein levels were assessed with a BCA kit (Epizyme Biotech, China). A total of 30–50 µg of soluble protein was loaded onto a 10–12.5% SDS–PAGE gel and electrotransferred onto a PVDF membrane (Millipore, Germany). The membranes were blocked at room temperature for 2 h using 5% nonfat milk (BioFroxx, China) in TBST. Primary antibodies (Table S1) were used to probe the protein blots overnight at 4 °C. Bands that reacted to the immune system were identified by incubating with HRP-linked secondary antibodies for 2 h. Then, the blots were visualized using a ChemiDoc XRS + imaging system (Bio-Rad), and the density of each band was analyzed with ImageJ software.

For immunoprecipitation, 10 µL of antibody was added to the samples at 4 °C for 6 h. After incubation, 25 µL of pretreated Protein A/G beads (YJ201; Epizyme, Shanghai, China) were added for an additional 12 h at 4 °C with rotation. The beads were subsequently washed three times with lysing buffer and eluted with sample buffer. Finally, SDS–PAGE electrophoresis was performed to separate the proteins following the same steps as those used for western blotting.

Statistical analysis

All the results are presented as mean ± SEM. We used IBM SPSS Statistics (v21.0; IBM, USA) to perform statistical analyses. The normality of the data distribution and homogeneity of variance were assessed using a Shapiro–Wilk test and Levene’s test, respectively. For comparisons between the two groups, either a two-tailed unpaired Student’s t test, Mann–Whitney U test, or Welch’s t test was applied, depending on the data distribution. For multigroup comparisons, one-way ANOVA followed by Tukey’s post hoc test was used when both normality and variance homogeneity were satisfied. When the data violated normality assumptions, a Kruskal–Wallis H test followed by Dunn’s post hoc test was performed. For data with unequal variances, Welch’s ANOVA followed by Games–Howell’s post hoc test was performed. A p value of less than 0.05 was considered to indicate statistical significance.

Results

HIBD induces endothelial cell pyroptosis and BBB disruption

First, we developed a rat model for HIBD (Fig. 1A). Left brain hemispheres (the injured side) were collected from the sham and HI groups and subjected to RNA sequencing analysis. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed notable enrichment in the “NOD-like receptor signaling pathway” when the HI group was compared to the sham group (Fig. 1B). The top 58 DEGs related to the NOD-like receptor signaling pathway are shown in Fig. 1C. Pyroptosis is widely recognized for its association with the activation of NOD-like receptors (NLRs), particularly the NLRP3 inflammasome, which is accompanied by GSDMD cleavage and the release of IL-18 and IL-1β [15]. To explore the occurrence of pyroptosis in HIBD, the levels of proteins linked to pyroptosis were compared between rats in the sham group and HI group. It was observed that rats in the HI group upregulated mRNA and protein levels of NLRP3, GSDMD, cleaved caspase-1, IL-18, ASC, and IL-1β (Fig. 1D–F). In addition, the ELISA results revealed a significant increase in IL-1β secretion from brain microvessel fractions after HIBD (Fig. 1G). Cell membrane impairment during pyroptosis may cause the release of cytoplasmic enzymes [39] such as lactate dehydrogenase (LDH), which was measured to quantify pyroptosis. Accordingly, LDH release was significantly greater in the HI group (Fig. 1H). To determine whether pyroptosis occurred in endothelial cells, CD31/caspase-1 immunofluorescence was performed. As shown in Fig. 1I, the level of caspase-1 was greatly increased in the endothelial cells of rats with HIBD. Thus, endothelial cell pyroptosis might contribute significantly to neonatal brain damage resulting from HI injury.

Fig. 1.

HI injury results in endothelial cell pyroptosis in rats. Postnatal day 7 rats were subjected to left common carotid artery ligation and hypoxia for 2.5 h, and brain tissues were isolated from sham and HI rats 24 h after injury. A Schematic diagram of the neonatal HIBD rat model. B Comparison of the top 20 KEGG pathways in the left brain hemisphere of the sham versus HI rats. C Hierarchical clustering of DEGs in the sham versus the HI group. Red signifies higher expression levels, and blue represents lower expression levels. D–F Pyroptosis-related protein levels (D), quantification (E), and mRNA levels (F) in rats (n = 5). G The level of IL-1β in brain microvessel fractions 24 h after HI injury was measured with an ELISA kit (n = 5). H LDH release was measured in rats following HI injury (n = 5). I Microphotographs of immunofluorescence staining and quantification of caspase-1 (green) colocalization in endothelial cells (CD31, red) 24 h after HI injury (n = 5; scale bar = 50 μm)

We also performed in vitro experiments to delve deeper into the function of pyroptosis in endothelial cells. HCMECs were subjected to OGD to simulate HI injury in vitro, and the effects were subsequently evaluated. Compared with that of the controls, cell survival decreased over time following OGD of varying durations (6–24 h), with 44.96 ± 1.41% survival at 18 h. Therefore, 18 h of OGD was used in subsequent experiments (Fig. S1A). Consistent with our in vivo data, OGD robustly upregulated the expression of pyroptosis biomarkers, such as NLRP3, GSDMD, cleaved caspase-1, ASC, IL-18, and IL-1β. This finding was consistent with the qRT–PCR (Fig. S1B) and western blotting results (Fig. S1C–D). Similarly, the intensity of capase-1 (Fig. S1E–F) and the release of LDH (Fig. S1G) were elevated in hCMECs after 18 h of OGD. Our findings indicated that endothelial cell pyroptosis occurred in rats and in hCMECs after HI injury or OGD.

To assess disruption of the BBB in the neonatal HIBD model, we measured the levels of junction proteins ZO-1, P120, β-catenin, and occludin 24 h after HI injury. Compared with those in the sham group, the levels of these junction proteins were dramatically lower in the HI group (Fig. 2A–B). Additionally, the permeability of the BBB was evaluated using Evans blue and FITC-dextran. We observed significant increases in Evans blue extravasation and FITC-dextran permeability in the HI group relative to the sham group (Fig. 2C–D). Similar to these in vivo results, OGD-treated hCMECs exhibited decreased levels of junction proteins (Fig. S2A–B) and ZO-1 fluorescence intensity (Fig. 2E). These cells also exhibited increased permeability when subjected to the FITC-dextran flux assay (Fig. 2F). Collectively, these findings indicated severe disruption of the BBB in HI-induced rats and OGD-treated hCMECs.

Fig. 2.

BBB integrity is impaired in rats and endothelial cells following injury. Brain tissues were isolated from sham and HI rats 24 h after injury; hCMECs were subjected to 18 h of OGD. A, B Protein levels (A) and quantitative analysis (B) of ZO-1, P120, β-catenin, and occludin 24 h after HI injury in rats (n = 5). C Images showing Evans blue-stained brains and a quantitative assessment of Evans blue fluorescence in the sham and HI groups (n = 5). D FITC-dextran extravasation in rats (n = 5). E Representative immunofluorescence images of ZO-1 (green) and quantification of ZO-1 expression in hCMECs using ImageJ software (n = 5; scale bar = 50 μm). F FITC-dextran extravasation in hCMECs (n = 5)

Activation of pyroptosis upregulates the expression of matrix metalloproteinases (MMPs) and exacerbates BBB impairment

Accumulating evidence indicates that pyroptosis regulates BBB integrity [40, 41]. We thus evaluated whether high levels of pyroptosis aggravated the loss of BBB integrity after HI injury. Nigericin, a pyroptosis-specific agonist, was intraperitoneally injected 2 h before HI injury. Interestingly, nigericin-mediated pyroptosis in HI-induced rats led to dramatic decreases in the expression of junction proteins, including ZO-1, P120, β-catenin, and occludin, as well as a decrease in the ZO-1 fluorescence signal (Fig. 3A–D). Similarly, HI injury increased Evans blue leakage and FITC-dextran permeability in rats, and these effects were aggravated by nigericin administration (Fig. 3E–G). In contrast, Ac-YVAD (a caspase-1 inhibitor) dramatically increased the expression of junction proteins in rats (Fig. S3A–B). We further investigated how nigericin promoted BBB disruption after HI induction. It is well known that MMPs degrade the extracellular matrix and components that connect cells; this is similar to the pathological processes seen in BBB disruption [42]. Therefore, the effects of nigericin on MMP-2 and MMP-9 activity were assessed. As shown in Fig. 3H, HI injury enhanced MMP-2/9 activity in rats, and nigericin administration further upregulated the expression of these proteins in the HI group compared with the sham group.

Fig. 3.

Pyroptosis disrupts BBB integrity by upregulating the expression of MMP2 and MMP9. Rats were intraperitoneally pretreated with the NLRP3 activator nigericin (4 mg/kg/day) 2 h before HI injury. A, B Protein levels (A) and quantitative analysis (B) of the junction proteins ZO-1, P120, β-catenin, and occludin in rats (n = 5). C, D Images (C) and quantitative analysis (D) of ZO-1 using double staining for ZO-1 (green) and CD31 (red) in rat brains (n = 5; scale bar = 50 μm). E, F Images of Evans blue-stained brains (F) and quantification (E) of Evans blue content in each group of rats (n = 5). G FITC-dextran extravasation in rats (n = 5). (H) Protein levels and quantification of MMP9 and MMP2 in rats (n = 5)

Consistent with the observed pyroptosis-related effects on MMPs and the BBB in vivo, the in vitro results revealed that nigericin promoted the expression of MMP-2 and MMP-9 (Fig. S4F–G) and increased BBB permeability in hCMECs (Fig. S4C). This was also demonstrated by elevated levels of MMP-2/9 (Fig. S4H–J) and decreased levels of ZO-1, P120, β-catenin, and occludin (Fig. S4A–B, S4D–E). Taken together, these findings suggest that HI-induced endothelial cell pyroptosis has a deleterious effect on the BBB.

HI injury increases the levels of PKM2

Pyroptosis is activated in endothelial cells following HI insult or OGD, but the underlying mechanism remains unclear. PKM2, a glycolytic enzyme, is an important regulator of glucose metabolism [23], and the role of PKM2 in endothelial cell metabolism has attracted much attention. It is believed that PKM2 also regulates endothelial cell survival and helps to maintain vascular integrity [43]. However, the relationship between PKM2 expression and endothelial cell function in HIBD remains unclear. We therefore used western blotting to investigate PKM2 levels in the brain after HI injury. PKM2 levels in the injured hemisphere started to increase at the first time point and peaked at 24 h (Fig. 4A). Moreover, in the cortex of the injured hemisphere (24 h after HI injury), PKM2 immunoreactivity increased in endothelial cells (CD31) (Fig. 4B). Similarly, OGD resulted in the upregulation of both PKM2 mRNA and protein levels in hCMECs (Fig. 4C–E).

Fig. 4.

PKM2 expression in endothelial cells from in vitro and in vivo HIBD models. A Western blot and quantification of the temporal profile of PKM2 levels in the left/injured hemisphere at 0, 6, 12, 24, and 48 h after HI injury (n = 5). B Immunofluorescence and quantitative analysis of PKM2 (green) colocalization on endothelial cells (CD31, red) 24 h after HI injury (n = 5; scale bar = 50 μm). C Western blot and quantification of PKM2 levels in hCMECs (n = 5). D qRT–PCR analysis of PKM2 expression in hCMECs (n = 5). E Immunofluorescence labeling and quantification of PKM2 (red) in hCMECs (n = 5)

PKM2 Inhibition triggers the activation of endothelial cell pyroptosis

To understand the relationship between PKM2 and pyroptosis in endothelial cells, we performed a series of rescue experiments using a PKM2 inhibitor (Compound 3k) and a pyroptosis inhibitor (MCC950). Intraperitoneal injections were administered 2 h before HI treatment, and comparisons were performed among the sham group, HI group, HI + Compound 3k group, and HI + Compound 3k + MCC950 group. Treatment with Compound 3k significantly increased endothelial cell pyroptosis, as demonstrated by elevated levels of NLRP3, GSDMD-N, cleaved caspase-1, ASC, IL-18, and IL-1β (Fig. 5A–B); increased caspase-1 immunofluorescence intensity (Fig. 5C–D); and a dramatic increase in LDH (Fig. 5E). However, MCC950 largely counteracted the effects of Compound 3k, resulting in the downregulation of pyroptosis-related proteins and LDH release.

Fig. 5.

PKM2 inhibition upregulates pyroptosis in HI-treated endothelial cells in vivo. Rats were intraperitoneally pretreated with the PKM2 inhibitor Compound 3k (5 mg/kg/day) alone or together with the NLRP3 inhibitor MCC950 (10 mg/kg/day) 2 h before HI injury. A, B Representative western blot bands (A) and quantitative analysis (B) of pyroptosis-related proteins in rats (n = 5). C, D Graphs of immunofluorescence staining (C) and quantitative analysis (D) of caspase-1 (green) colocalization on endothelial cells (CD31, red) in rats (n = 5; scale bar = 50 μm). E LDH release was assessed in rats after HI injury (n = 5)

In parallel, to explore the function of PKM2 in controlling pyroptosis in endothelial cells, we used PKM2 siRNA to reduce PKM2 expression in hCMECs. Following knockdown of PKM2, endothelial cell pyroptosis was largely promoted compared with that in the control siRNA groups (as indicated by the upregulation of proteins involved in pyroptosis). Importantly, the effects of PKM2 knockdown were largely reversed by the pyroptosis inhibitor MCC950 (Fig. S5A–D), which indicates that decreased PKM2 activity in endothelial cells aggravated pyroptosis.

PKM2 activation alleviates HI-induced brain injury and BBB dysfunction

To evaluate the effects of PKM2 on brain damage and BBB function after HI injury, SD rats were intraperitoneally injected with a PKM2 activator (ML265) 2 h before HIBD. TTC staining was subsequently performed on the brain sections to identify differences between living and dead tissue, and subsequent analyses revealed that ML265 dramatically reduced the brain infarct volume (29.28 ± 1.44%) compared with that in the HI group (49.51 ± 4.08%) (Fig. 6A). Brain damage was quantified by determining the degree of loss in volume at 7 d after HIBD. H&E staining of treated brains confirmed that the HI + ML265 group exhibited a lower loss of brain tissue than the HI group (26.40 ± 1.81% vs. 47.60 ± 2.75%) (Fig. 6B–C). To evaluate neurological function, an MWM test was conducted on rats 21 d following HIBD. There was no difference in swimming speed among the groups (Fig. 6D). In all the learning and memory tests, compared with the sham group, the HIBD group performed poorly and exhibited longer mean escape latency (43.53 ± 3.34 s vs. 7.82 ± 1.61 s, p < 0.001) (Fig. 6E–F) and lower crossing frequency (0.75 ± 0.25 vs. 3.00 ± 0.27, p < 0.001) (Fig. 6G–H). Importantly, ML265 treatment (HI + ML265 group) significantly improved these neurological deficits compared to the HI group.

Fig. 6.

PKM2 activation can attenuate HI-induced brain injury and BBB disruption. Rats were given the PKM2 activator ML265 (20 mg/kg/day) intraperitoneally 2 h before HI injury and every 24 h thereafter. A Representative coronal sections stained with TTC and quantification of infarct size 24 h after HI injury in each group (n = 8). B Quantitative analysis of cerebral tissue loss (n = 5). C Coronal brain images stained with H&E 7 d after HI (n = 5). D Images showing the swimming speed of rats in different groups (n = 8). E Images showing the swimming paths of rats in different groups to the location of the hidden platform. F Mean escape latency in finding the hidden platform in the different groups (n = 8). G Representative swim route traces for rats after withdrawing the hidden platform on the sixth day of the MWM test. H The percentage of rats crossing the initial platform location during 60 s trials (n = 8). I Representative whole-brain images depicting Evans blue leakage and measurement of the Evans blue concentration in lesions from each group (n = 5). J FITC-dextran extravasation in rats following HI injury (n = 5)

We further explored the role of PKM2 activation in regulating BBB function. The PKM2 activator ML265 was used to upregulate PKM2 in rats and hCMECs. In the presence of ML265, HI-induced BBB impairment was largely ameliorated, as indicated by decreased Evans blue extravasation and FITC-dextran permeability (Fig. 6I–J); we also observed dramatically increased levels of junction proteins (ZO-1, P120, β-catenin, and occludin) (Fig. 7A–B) and increased ZO-1 immunofluorescence intensity (Fig. 7C–D). In parallel, OGD-induced BBB impairment was strongly ameliorated by the addition of ML265 to hCMECs, as indicated by increased levels of junction proteins (Fig. 7E–G) and reduced FITC-dextran permeability (Fig. 7H). These results indicate that the upregulation of PKM2 expression ameliorated brain injury and restored BBB function in HIBD rats and hCMECs.

Fig. 7.

PKM2 activation restores BBB integrity in vivo and in vitro. Rats were intraperitoneally pretreated with the PKM2 activator ML265 (20 mg/kg/day) 2 h before HI injury; hCMECs were pretreated with ML265 (20 µM) for 2 h, followed by OGD for an additional 18 h. A, B Protein levels (A) and quantification (B) of ZO-1, P120, β-catenin, and occludin 24 h after HI injury in rats (n = 5). C, D Representative images of immunofluorescence staining (C) and quantitative analysis (D) of ZO-1 (green) colocalization on endothelial cells (CD31, red) in rats (n = 5; scale bar = 50 μm). E–G Western blot (E, F) and quantification (G) of junction proteins (ZO-1, P120, β-catenin, and occludin) in hCMECs (n = 5). H FITC-dextran extravasation in hCMECs following OGD (n = 5)

PKM2 preconditioning contributes to the production of lactate and ROS

Accumulating evidence suggests that endothelial cell metabolism is involved in various diseases [44, 45]. To verify that PKM2 affects endothelial cell metabolism, we used the pcDNA 3.1-PKM2 vector to overexpress PKM2 in hCMECs. First, we assessed the levels of lactate, a product of glycolysis. Compared with the control, overexpression of PKM2 in hCMECs resulted in elevated levels of lactate (Fig. 8A). Evidence also suggests that lactate promotes cell survival, and increased levels of lactate have been linked to oxidative stress defense [46, 47]. We therefore assessed the effect of lactate production upon ROS release in the presence of the LDH blocker galloflavin. The overexpression of PKM2 promoted a mild ROS burst in hCMECs, which was reversed by galloflavin intervention. Furthermore, slightly higher levels of ROS, which were promoted by PKM2, contributed to a reduced level of ROS after OGD. However, cotreatment with galloflavin under OGD conditions counteracted the protective effects of PKM2 preconditioning, which increased ROS levels (Fig. 8B–C). Furthermore, overexpression of PKM2 inhibited 8-OHdG (an oxidative stress marker) expression after OGD (Fig. 8D). A similar attenuation of the effect of PKM2 was also observed when oxamate, another LDH inhibitor, was used. Conversely, pretreatment with lactate counteracted an increase in ROS levels induced by oxamate treatment (Fig. S6). These analyses revealed that preconditioning with PKM2 slightly increased the levels of lactate and ROS and inhibited oxidative stress in control hCMECs, and preconditioning decreased the levels of lactate and ROS in OGD-treated hCMECs.

Fig. 8.

PKM2 overexpression promotes lactate and ROS generation, which can be counteracted by LDHi. HCMECs were pretreated with LDHi (Galloflavin, 15 µM) for 2 h alone or together with recombinant plasmid Flag-PKM2 for 6 h, followed by OGD for an additional 18 h. A Quantification of lactate levels in hCMECs with OGD or PKM2 overexpression (n = 5). B, C Quantification (B) and representative images (C) of the results of the ROS staining assay (n = 5; scale bars = 100 μm). D Quantification and representative images of the 8-OHdG staining assay (n = 3; scale bars = 50 μm)

Activation of PKM2 suppresses endothelial cell pyroptosis via the NRF2/TRX/TXNIP axis

It has been reported that low levels of ROS signaling activate the antioxidant system (including NRF2) to counteract damage induced by OGD [48]. To elucidate the effects of PKM2 on the downstream antioxidant system, PKM2 was overexpressed in hCMECs subjected to OGD, and the levels of NRF2, thioredoxin (TRX), and HO-1 were assessed. Antioxidative indicators (NRF2, TRX, and HO-1) were upregulated, whereas pyroptosis-related proteins (NLRP3 and TXNIP) were notably inhibited by PKM2 overexpression. Importantly, the effects of PKM2 overexpression were largely reversed by the NRF2 inhibitor ML385 (Fig. 9A, E, G). Furthermore, western blot analysis of nuclear proteins revealed that PKM2 overexpression increased NRF2 expression in the nucleus in the control group relative to the OGD group. However, cotreatment with ML385 during OGD almost entirely prevented the nuclear localization of NRF2 induced by PKM2 overexpression in hCMECs (Fig. 9B, C). Immunofluorescence staining of NRF-2 and HO-1 in hCMECs revealed similar effects on NRF2 nuclear translocation and HO-1 expression (Fig. 9F, H–J). In addition, we verified the interaction between NLRP3 and TXNIP by coimmunoprecipitation in hCMECs (Fig. 9D).

Fig. 9.

PKM2 overexpression promotes NRF2 expression to inhibit TXNIP and pyroptosis in hCMECs. HCMECs were pretreated with the NRF2 inhibitor ML385 (20 µM) for 2 h alone or together with the recombinant plasmid Flag-PKM2 for 6 h, followed by OGD for an additional 18 h. A, E, G Protein levels (A) and quantification (E, G) of NRF2 in the cytoplasm and TXNIP, HO-1, TRX, and NLRP3 in hCMECs (n = 5). B, C Western blot (B) and quantification (C) for nuclear NRF2 in hCMECs (n = 5). D Immunoprecipitation with anti-NLRP3 or anti-TXNIP from control and OGD-treated hCMECs. F, I Immunofluorescence images (F) and quantification (I) of NRF2 expression 18 h after OGD (n = 5; scale bar = 50 μm). H, J Immunofluorescence images (H) and quantification (J) of HO-1 expression 18 h after OGD (n = 5; scale bar = 50 μm)

We then validated these results in vivo. HI-induced oxidative stress in rats was strongly altered by ML265 (a PKM2 activator), as indicated by the upregulation of NRF2, TRX, and HO-1 and the downregulation of NLRP3, 4-HNE, and TXNIP. Furthermore, ML385 largely counteracted decreases in NLRP3, 4-HNE, and TXNIP expression and increases in NRF2, TRX, and HO-1 expression (Fig. S7A–D). Immunoprecipitation experiments revealed that ML265 treatment substantially suppressed the interaction between NLRP3 and TXNIP in HI rats, whereas the administration of Compound 3k reversed this effect (Fig. S7E). In addition, the ROS scavenger NAC largely inhibited the effects of ML265 on the NRF2/TRX/TXNIP axis and pyroptosis (Fig. S8A–D). These data indicate that PKM2 preconditioning increased NRF2, 4-HNE, and TRX expression, thus inhibiting TXNIP and pyroptosis, which depend on ROS generation.

Discussion

Our findings reveal a previously unrecognized mechanism by which PKM2 preconditioning enhances endothelial cell survival and preserves BBB integrity in neonatal HIBD. As a central component of the neurovascular unit, cerebral endothelial cells rely heavily on glycolysis for energy production and barrier maintenance [43, 49]. Yet, how endothelial cell metabolism contributes to HIBD pathology has remained unclear. In this study, disrupting glycolysis, either by endothelial PKM2 knockdown or systemic administration of Compound 3k, exacerbated endothelial pyroptosis, BBB breakdown, and brain injury. In contrast, PKM2 preconditioning through PKM2 overexpression or ML265 treatment protected endothelial cells and attenuated BBB disruption. Mechanistically, PKM2-driven glycolytic activity increased lactate production and induced a controlled ROS elevation, which subsequently activated the NRF2/TRX/TXNIP axis to restrict pyroptosis and promote vascular stability.

PKM2 exists in two forms: dimeric and tetrameric. The tetramer serves mainly as a pyruvate kinase and regulates glycolysis in the cytoplasm, whereas the dimer can enter the nucleus to regulate gene expression [23, 50]. PKM2 exhibits distinct functional disparities in endothelial cells, neurons, and microglia because of their differing metabolic requirements and cellular functions [51, 52]. In endothelial cells, which are vital for blood vessel function and angiogenesis, PKM2 regulates glycolysis, ATP production, and cell junctions [53]. In contrast, in neurons and microglia, PKM2 mainly regulates gene expression in the nucleus, affecting inflammation and cell death [54, 55]. Recent studies have shown that PKM2 has neuroprotective effects [56]. Gao et al. (2024) reported that in microglia, inhibiting the ability of nuclear PKM2 to modify histone H3/HIF-1α resulted in anti-neuroinflammatory effects [57]. PKM2 also protects against neurodegeneration via the PKM2/NRF2/ARE axis by alleviating oxidative stress in neurons [56]. Moreover, the biological effects of PKM2 in neutrophils and immune cells appear to be largely mediated by nuclear effects (e.g., via STAT3 or HIF-1α) [25, 58]. Our findings indicate that overexpression of PKM2 or using ML265 protect endothelial cells from pyroptosis and preserve BBB function. This result differs from findings in microglia, neurons, and immune cells, which may be due to endothelial cells relying more on glycolysis and PKM2’s pyruvate kinase activity [28]. PKM2 plays a crucial role in regulating cellular metabolism in endothelial cells, with its activity and function determined by the balance between its dimeric and tetrameric forms [59]. This balance is important for angiogenesis, endothelial cell junction dynamics, and overall vascular function [60]. Ren et al. (2020) reported that PKM2 can modulate glycolysis and mitochondrial fission and fusion, thereby influencing angiogenesis [53]. In addition, PKM2-driven lactate overproduction can trigger an endothelial-to-mesenchymal transition, thus contributing to fibrosis and necrosis under ischemic conditions [60]. Furthermore, proper ATP generation, which is facilitated by tetrameric PKM2, is needed for maintaining barrier integrity and allowing collective cell migration [51]. Thus, understanding how PKM2’s forms are regulated could offer new insights into treating vascular and metabolic diseases.

Lactate, a product of glycolysis, is now receiving a great deal of attention for its role as a signaling agent in maintaining cell survival and BBB integrity. Tauffenberger et al. (2019) reported that lactate increased the survival of SH-SY5Y cells by activating PI3K signaling and regulating protein homeostasis in the endoplasmic reticulum [47]. Endothelium-derived lactate is also required for pericyte function (via glucose transporter-1) to maintain BBB integrity [49]. Consistent with previous reports [47, 61], our research revealed that PKM2 overexpression increased lactate levels and induced a mild ROS response in hCMECs (Fig. 8A-C). Under conditions of oxidative stress, ROS react with and modify specific cysteine residues of KEAP1, which prevents KEAP1 from ubiquitinating NRF2 and allows NRF2 to move to the nucleus, where it attaches to AREs and triggers the activation of target genes (including the TRX antioxidant system) [62]. The TRX system, which is crucial for maintaining cellular redox balance, includes TRX, TRXR, TXNIP, and NADPH [63]. Moreover, TXNIP, a redox-sensitive protein, has the ability to bind to and activate the NLRP3 inflammasome. Increased binding of TXNIP and TRX and reduced TXNIP/NLRP3 interaction negatively regulate pyroptosis [64]. The reversal of these protective effects by the NRF2 inhibitor ML385 further supports this mechanism.

Although pyroptosis has been documented in neurons, microglia and astrocytes during HIBD [17, 18, 65], its role in endothelial cells has been unclear. Our in vivo and in vitro models revealed marked pyroptotic activation in endothelial cells following HI or OGD injury. Additionally, we revealed BBB disruption in neonatal rats following HI damage including severe brain edema, reduced tight/adherens junction proteins, and increased leakage of Evans blue and FITC-dextran. Importantly, inhibition of pyroptosis via MCC950 ameliorated BBB damage, confirming a causal relationship. Pyroptosis also increased MMP-2/9 expression (Fig. 3H, S4F), further weakening the BBB by degrading junctional proteins, consistent with our previous findings [66–68]. Interestingly, upregulation of these MMPs was further elevated by the pyroptosis agonist, nigericin. Together, these results position endothelial pyroptosis as a central driver of BBB breakdown in neonatal HIBD.

Several limitations remain to be addressed in further studies. (i) The administration of systemic pharmacological inhibitors/activators may have off-target effects. Therefore, the observed protective effects cannot be attributed exclusively to endothelial PKM2, as other cell types such as neurons, astrocytes, and microglia may also respond to these interventions. Future work employing endothelial-specific knockout or endothelial-specific adeno-associated viruses will be necessary to determine whether the effects observed in vivo are mediated directly by endothelial cells. (ii) Our in vivo experiments were performed using male pups only due to sex-differentiated responses to oxidative stress. Our findings may not fully extend to female animals. Future experiments including both sexes will be necessary to determine whether the observed effects are sex-specific. (iii) Although our in vitro experiments provide mechanistic insights, the use of immortalized cell lines represents an inherent limitation. In particular, primary rat brain microvascular endothelial cells (BMECs) would more faithfully preserve the native endothelial phenotype, tight-junction characteristics, and cell–cell interactions that are critical for accurately assessing PKM2-dependent vascular responses. Therefore, future studies incorporating primary BMECs, or ideally co-culture systems with neurons and glia, will help validate our findings and better recapitulate the in vivo microenvironment. (iv) While our data indicate that PKM2-driven lactate contributes to mild ROS generation and activation of NRF2/TRX/TXNIP signaling, we did not include monocarboxylate transporter (MCT) inhibitors to block lactate transport. Future experiments using pharmacological or genetic inhibition of MCTs will be necessary to confirm the causal role of extracellular lactate uptake in endothelial protection.

Conclusion

In summary, we showed that PKM2 preconditioning attenuated endothelial cell pyroptosis and BBB disruption in hCMECs subjected to OGD and in a Rice–Vannucci rat model. In both systems, this protective mechanism was mediated by oxidative stress resistance and the NRF2/TRX/TXNIP signaling pathway; PKM2 preconditioning induced lactate production and promoted ROS-mediated oxidative stress defense mechanisms in HIBD (Fig. 10). Thus, this research improves our mechanistic understanding of HIBD and provides a foundation for future drug development that targets PKM2 to treat neonatal HIBD.

Fig. 10.

Schematic showing that PKM2 preconditioning improves HIBD through oxidative stress resistance and activation of NRF2/TRX/TXNIP signaling, which inhibits endothelial cell pyroptosis and alleviates BBB disruption

Authors’ contributions

Yingying Hu was involved in designing the research, conducting the experiments, analyzing the data, and writing the manuscript. Guosheng Yu performed bioinformatic analysis. Liying Lu, Yan Nan, Xinyi Wang and Yani Feng participated in the experiments. Jianghu Zhu helped to data acquisition. Xingyun Wang and Zhenlang Lin designed the study and revise the article. All authors read and agreed to the published version.

Funding

This work was supported by grants from National Natural Science Foundation of China (82271747) and Key Research and Development Program of Zhejiang Province (2025C02081).

Data availability

The transcriptome sequencing data are available in the BioSample database (BioProject ID: PRJNA1163730).

Declarations

Ethics approval and consent to participate

This study’s animal care and experiments adhered to the established guidelines for the Care and Use of Laboratory Animals of the National Institute of Health, and was authorized by the Experimental Animal Ethics Committee of Wenzhou Medical University (ethics NO. wydw2023-0559).

Competing interests

The authors declare no competing interests.